MPC in Power Electronics and Drives: 40 Years of Research, Recent Industrial Successes and a Path Forward-Slides

Abstract:This lecture explores the exciting history and evolution of model predictive control (MPC) in power electronics and drives. MPC was conceived in the process industry in the 1970s to optimally control highly nonlinear systems with hundreds of state variables and numerous hard constraints. The basic MPC principle was re-imagined in power electronics in the 1980s. Finite control set MPC, which is typically formulated as a discrete deadbeat controller, followed in the early 2000s. Recently, advanced MPC methods have started to revolutionize high-power converters by increasing their rated power, lowering their cost and ensuring their safe operation in the presence of grid disturbances and faults. In particular model predictive pulse pattern control (MP3C) fully exploits the performance potential of high-power converters by combining the benefits of MPC with the optimal steady-state performance of optimized pulse patterns. For load commutated inverters, model predictive torque control (MPTC) increases the drive's efficiency and robustness to grid disturbances. The lecture concludes with an assessment of the MPC methods available today. The necessity for sophisticated MPC methods that maximize the converter and drive performance will be discussed, and exciting research opportunities for academics and business opportunities for industry will be pointed out.